Dissertation Announcement for Chenxu Zhang – 05/03/2022 at 2:00 PM

April 20, 2022

Dissertation Title: Short-term electricity price point and probabilistic forecasts

When: Tuesday, May 3, 2022, 2:00 PM to 4:00 PM

Where: In person-Simrall-228 (Conference room) Remote access: https://msstate.webex.com/msstate/j.php?MTID=m96faf63c7396d2a56aa1e6fde81a74f0

Candidate: Chenxu Zhang

Degree: Doctor of Philosophy, Electrical and Computer Engineering

Committee Members:
Dr. Yong Fu
(Major Professor)

Dr. Masoud Karimi
(Committee Member)

Dr. Seungdeog Choi
(Committee Member)

Dr. Bo Tang
(Committee Member)

Accurate short-term electricity price forecasts are essential to all electricity market participants. Generation companies adopt price forecasts to hedge generation shortage risks; for load serving entities, accurate price forecasts assist them to get sufficient amount of energy with low cost; also, trading companies utilize price forecasts to arbitrage between the day-ahead and the real-time market.

Currently, researches on point forecast mainly focus on exploring the periodic patterns of electricity price in time domain. However, few studies analyze electricity price periodicity in frequency domain that is able to identify more information within price data to facilitate forecast. Besides, the price spike forecast has not been fully studied in the existing works. Therefore, we propose a short-term electricity price forecast framework that analyzes price data in frequency domain and takes price spike prediction into consideration. First, the variational mode decomposition (VMD) is adopted to decompose price data into multiple frequency band-limited modes where each mode has a central frequency. Then, the extended discrete Fourier transform (EDFT) is adopted to transform the decomposed price mode into frequency domain and perform normal price forecasts. In addition, we utilize the enhanced structure preserving oversampling (ESPO) and synthetic minority oversampling technique (SMO
TE) to oversample price spike cases to improve price spike forecast accuracy.

In addition to point forecasts, market participants also need probabilistic forecasts to quantify uncertainties within predictions. However, there are several shortcomings within current researches. The first one relates to probabilistic forecasts quality. Although wide prediction interval satisfies reliability requirement, the over-width intervals incur market participants to derive conservative decisions that result in benefit reduction. The second one is about the probabilistic forecast assumption. Although electricity price data follow heteroscedasticity distribution, to reduce computation burden, many researchers assume that price data follow the normal distribution. However, such assumption under different market conditions is still questionable. Therefore, to handle the above-mentioned deficiencies, we propose an optimal prediction interval method considering both reliability and sharpness requirements and without assuming the distribution of electricity prices. 1) By taking b oth reliability and sharpness into consideration, we ensure the prediction interval has a narrow width without sacrificing reliability. 2) To avoid electricity prices distribution assumptions, we utilize the quantile regression to estimate the upper and lower bound of prediction interval. 3) Exploiting the versatile and fast-training abilities, the extreme learning machine (ELM) method is adopted to forecast prediction intervals.

The effectiveness of our proposed point and probabilistic forecast methods are justified by using the actual price data from various electricity markets. Comparing with the predictions derived from others’ researches, numerical results show that our methods could provide accurate and stable forecast results under different market situations.